Samsung SDS America’s Cloud Native Computing Team recently ran Navops Launch through its paces to see how it compared to provisioning a bare metal Kubernetes cluster by hand. They started with a 20-node un-provisioned compute cluster and some ready-to-run Docker containers. Let’s take a look at how Navops Launch stacked up (pun intended) versus the manual installation method.
The objective of the comparison was to quantitatively measure the Navops Launch installation process and to compare this to the community-defined steps to deploy a container cluster using the common stack of tools.
We are pleased with the outcome of these tests and have summarized the finding below, and have also included detailed results from Samsung SDS America’s report directly.
Quick Results Summary:
Let’s take a closer look at how Launch measured up. The content below is extracted directly from Samsung SDS America’s report:
The graphic below provides a comparison of key installation metrics between the two installation approaches:
Note: The 1 Node installation times were significantly higher than multi-node clusters because initial learning and training was required for the systems administrator. This was done intentionally to measure the required ramp-up time.
From the perspective of a sysadmin, we judged Navops Launch as being very successful.
From the perspective of the application developer, not strictly a devops related perspective, we again judged Navops as being very successful, but for one very different reason.
Taken together, these factors all lend themselves to a very positive user experience.
Want to build a Container Cluster yourself; Navops Launch is available now for free download at www.navops.io.
 Univa works with Samsung SDSA to leverage their participation in the Kubernetes community efforts, and to take advantage of their service deployment and development experience. Although Samsung SDSA cannot endorse third party products, this report represents their unbiased testing results, as reported to Univa.